Biomedical diagnosis based on ion mobility spectrometry - A case study using probabilistic relational modelling and learning

Abstract

Aiming at providing a non-invasive and easy-to-use method for the early detection of bronchial carcinoma, it has been proposed to apply ion mobility spectrometry (IMS) to the breath a person exhales. Extending previous work using such IMS data, we report on a case study using methods of probabilistic relational modelling and learning. By taking additional features of an IMS measurement into account and using refined clustering and modelling methods, inference accuracy is increased. © 2012 Springer-Verlag.